public void Convert(PhysicalInstance pi)
{
if (pi == null)
{
throw new ArgumentNullException(nameof(pi));
}
if (pi.RecordLayouts.Count == 0)
{
return;
}
// Create a single DataRelationship for the PhysicalInstance.
// Within this, we will add one LogicalRecord per old RecordLayout.
var dataRelationship = new DataRelationship()
{
AgencyId = pi.AgencyId
};
pi.DataRelationships.Add(dataRelationship);
// For each record layout, add the variables to a new DataRelationship.
foreach (var recordLayout in pi.RecordLayouts)
{
// Create a LogicalRecord with the same descriptive information
// as the RecordLayout we are migrating.
var logicalRecord = new LogicalRecord {
AgencyId = pi.AgencyId
};
dataRelationship.LogicalRecords.Add(logicalRecord);
logicalRecord.ItemName.Copy(recordLayout.ItemName);
logicalRecord.Label.Copy(recordLayout.Label);
logicalRecord.Description.Copy(recordLayout.Description);
// For each DataItem in the RecordLayout, add a VariablesInRecord entry
// to the LogicalRecord.
foreach (var dataItem in recordLayout.DataItems)
{
logicalRecord.VariablesInRecord.Add(dataItem.Variable);
}
}
// Remove the old RecordLayouts.
pi.RecordLayouts.Clear();
}
public ResourcePackage Import(string path, string agencyId)
{
this.harmonizingCache = new HarmonizingCache(MultilingualString.CurrentCulture);
var resourcePackage = new ResourcePackage();
resourcePackage.AgencyId = agencyId;
logger.Debug("Importing RData");
if (string.IsNullOrEmpty(path))
{
throw new ArgumentNullException("fileName");
}
if (!File.Exists(path))
{
throw new ArgumentException("The specified file must exist");
}
string fileNameWithExtension = Path.GetFileName(path);
string fileNameOnly = Path.GetFileNameWithoutExtension(path);
logger.Debug("RData import file: " + fileNameOnly);
resourcePackage.DublinCoreMetadata.Title.Current = fileNameOnly;
// Create the PhysicalInstance.
var physicalInstance = new PhysicalInstance()
{
AgencyId = agencyId
};
resourcePackage.PhysicalInstances.Add(physicalInstance);
physicalInstance.DublinCoreMetadata.Title.Current = fileNameOnly;
// File location
if (path != null)
{
DataFileIdentification fileID = new DataFileIdentification();
Uri uri;
if (Uri.TryCreate(path, UriKind.RelativeOrAbsolute, out uri))
{
fileID.Uri = uri;
}
fileID.Path = path;
physicalInstance.FileIdentifications.Add(fileID);
}
// Create the DataRelationship.
var dataRelationship = new DataRelationship();
physicalInstance.DataRelationships.Add(dataRelationship);
dataRelationship.AgencyId = agencyId;
dataRelationship.Label.Current = fileNameOnly;
// Load the file into R.
string pathForR = path.Replace("\\", "/");
engine.Evaluate(string.Format("load('{0}')", pathForR));
// Find all the data frames.
var dataFrames = GetDataFrames();
// For each data frame in the RData file, create a LogicalRecord.
foreach (var pair in dataFrames)
{
string name = pair.Key;
var dataFrame = pair.Value;
// TODO This should be tracked per record, not PhysicalInstance.
physicalInstance.FileStructure.CaseQuantity = dataFrame.RowCount;
var logicalRecord = new LogicalRecord()
{
AgencyId = agencyId
};
dataRelationship.LogicalRecords.Add(logicalRecord);
logicalRecord.Label.Current = name;
List <string> variableLabels = null;
var variableLabelsExpr = dataFrame.GetAttribute("var.labels");
if (variableLabelsExpr != null)
{
var labelVector = variableLabelsExpr.AsVector();
variableLabels = new List <string>(labelVector.Select(x => (string)x));
}
for (int i = 0; i < dataFrame.ColumnCount; i++)
{
string columnName = dataFrame.ColumnNames[i];
var column = dataFrame[i];
var variable = new Variable()
{
AgencyId = agencyId
};
logicalRecord.VariablesInRecord.Add(variable);
// Name
variable.ItemName.Current = columnName;
// Label
if (variableLabels != null)
{
variable.Label.Current = variableLabels[i];
}
// Type
if (column.Type == RDotNet.Internals.SymbolicExpressionType.NumericVector)
{
variable.RepresentationType = RepresentationType.Numeric;
variable.Additivity = AdditivityType.Stock;
}
else if (column.Type == RDotNet.Internals.SymbolicExpressionType.IntegerVector)
{
if (column.IsFactor())
{
variable.RepresentationType = RepresentationType.Code;
string[] factors = column.AsFactor().GetLevels();
variable.CodeRepresentation.Codes = GetCodeList(factors, agencyId, resourcePackage);
}
else
{
variable.RepresentationType = RepresentationType.Numeric;
variable.NumericRepresentation.NumericType = NumericType.Integer;
variable.Additivity = AdditivityType.Stock;
}
}
else if (column.Type == RDotNet.Internals.SymbolicExpressionType.CharacterVector)
{
variable.RepresentationType = RepresentationType.Text;
}
}
}
return(resourcePackage);
}
